Files
group_fqcd_jr/tools/check_portal_modules.py
T

254 lines
10 KiB
Python
Raw Normal View History

"""验证投顾工作台的前端模块:语法正确、import 能解析、用到的符号有来源。
## 为什么要这个检查
2026-09-13 合并组员提交后发现:`dashboard.js` 被拆成三个模块后,
**`CONTENT_TYPE_LABELS` 的定义搬走了、import 却忘了加** —— 文件里还在用这个常量。
这类问题单元测试抓不到(前端 js 不过 Python 的导入检查),
只会在浏览器里以 `ReferenceError: CONTENT_TYPE_LABELS is not defined` 的形式爆出来,
表现为「投顾工作台打开是白板」,而且**测试全绿**。
所以这里做四件事:
1. 每个文件交给 `node --check` 按 ES module 解析(语法错当场暴露);
2. 每个相对 import 的目标文件必须存在;
3. 每个 import 进来的名字,在目标文件里必须真的有 `export`;
4. 文件里用到的全大写常量(`FOO_BAR` 形态、且不是 `x.FOO_BAR` 属性访问),
必须在「import 进来的 + 本文件声明的 + 已知全局」里 —— 第 4 条正是抓上面那个 bug 的。
## 第 4 条的取词范围(`code_only()`)
只保留**真代码**:注释、`'…'` / `"…"` 字符串、以及模板字符串里的**文字部分**都被清空,
唯独 `${…}` 里的表达式保留下来(那里是真代码,也正是原 bug 容易藏身的地方)。
早期版本只剔除引号字符串,结果注释里的 `HHI`、文案里的 `AUM`、模板串里的 `QDII`
全被当成"未声明的变量" —— 误报比真问题还多,检查会被当成摆设跳过。
"""
from __future__ import annotations
import re
import shutil
import subprocess
import sys
import tempfile
from pathlib import Path
PROJECT_ROOT = Path(__file__).resolve().parents[1]
PORTAL = PROJECT_ROOT / "app" / "static" / "portal"
DASHBOARD_DIR = PORTAL / "employee-advisor" / "dashboard"
MODULES = [
"dashboard.js",
"actions-module.js",
"history-module.js",
"service-request-module.js",
"published-module.js",
"assistant-module.js",
"customer-module.js",
"advisor-engine.js",
"advisor-config.js",
]
#: 浏览器与运行时天然存在的名字,不算"没来源"。
KNOWN_GLOBALS = {
"JSON", "Math", "Number", "String", "Boolean", "Object", "Array", "Promise",
"Date", "Error", "Map", "Set", "RegExp", "Symbol", "URL", "URLSearchParams",
"NaN", "Infinity", "undefined", "null", "true", "false", "console", "window",
"document", "localStorage", "sessionStorage", "fetch", "setTimeout",
"clearTimeout", "setInterval", "clearInterval", "requestAnimationFrame",
"HTMLElement", "Node", "Event", "CustomEvent", "AbortController",
}
IMPORT_RE = re.compile(
r"import\s+(?:(?P<names>\{[^}]*\})|(?P<default>[\w$]+))\s+from\s+['\"](?P<src>[^'\"]+)['\"]"
)
EXPORT_RE = re.compile(
r"export\s+(?:const|let|var|function|class)\s+(?P<name>[\w$]+)"
r"|export\s*\{(?P<list>[^}]*)\}"
)
#: 全大写常量:至少两个字符、全大写/数字/下划线,且不是"前面带点"的属性访问。
CONST_USE_RE = re.compile(r"(?<![\w.$])([A-Z][A-Z0-9_]{2,})\b")
#: 对象字面量里作为 key 的大写标识符(`{ FOO: 1 }`)不算使用。
OBJECT_KEY_RE = re.compile(r"(?<![\w.$])([A-Z][A-Z0-9_]{2,})\s*:")
def code_only(source: str) -> str:
"""只保留真代码:清空注释、引号字符串与模板字符串的文字部分(`${…}` 里的表达式保留)。
逐字符扫描而不是一把正则:模板字符串里可以嵌套 `${…}`(甚至再嵌模板),
正则匹配不了嵌套,只能老老实实走一遍。
"""
out: list[str] = []
index = 0
length = len(source)
while index < length:
char = source[index]
following = source[index + 1] if index + 1 < length else ""
if char == "/" and following == "*":
end = source.find("*/", index + 2)
index = length if end == -1 else end + 2
out.append(" ")
continue
if char == "/" and following == "/":
end = source.find("\n", index)
index = length if end == -1 else end
out.append(" ")
continue
if char in "'\"":
cursor = index + 1
while cursor < length:
if source[cursor] == "\\":
cursor += 2
continue
if source[cursor] == char:
break
cursor += 1
out.append(" ")
index = min(cursor + 1, length)
continue
if char == "`":
index += 1
while index < length:
if source[index] == "\\":
index += 2
continue
if source[index] == "`":
index += 1
break
if source[index] == "$" and index + 1 < length and source[index + 1] == "{":
depth = 1
index += 2
start = index
while index < length and depth:
if source[index] == "{":
depth += 1
elif source[index] == "}":
depth -= 1
if depth == 0:
break
index += 1
out.append(" " + source[start:index] + " ")
index += 1
continue
index += 1
out.append(" ")
continue
out.append(char)
index += 1
return "".join(out)
def exported_names(path: Path) -> set[str]:
source = path.read_text(encoding="utf-8")
names: set[str] = set()
for match in EXPORT_RE.finditer(source):
if match.group("name"):
names.add(match.group("name"))
if match.group("list"):
for piece in match.group("list").split(","):
piece = piece.strip()
if not piece:
continue
names.add(piece.split(" as ")[-1].strip())
return names
def resolve_import(source: Path, spec: str) -> Path | None:
"""相对 import 才能解析到本地文件;`/static/...` 这类站内绝对路径跳过。"""
if not spec.startswith("."):
return None
target = (source.parent / spec.split("?")[0]).resolve()
return target if target.is_file() else None
def check_syntax(paths: list[Path]) -> list[str]:
"""用 node 按 ES module 解析。装不了 node 就跳过(不让环境问题变成假失败)。"""
node = shutil.which("node")
if node is None:
return []
problems: list[str] = []
with tempfile.TemporaryDirectory() as tmp:
for path in paths:
# `node --check` 按扩展名决定模块类型,所以复制成 .mjs 再检查。
probe = Path(tmp) / (path.stem + ".mjs")
probe.write_text(path.read_text(encoding="utf-8"), encoding="utf-8")
result = subprocess.run(
[node, "--check", str(probe)], capture_output=True, text=True, check=False
)
if result.returncode != 0:
tail = (result.stderr or "").strip().splitlines()
problems.append(f"{path.name}: 语法错误 {tail[-1] if tail else ''}")
return problems
def main() -> int:
paths = [DASHBOARD_DIR / name for name in MODULES]
missing = [p.name for p in paths if not p.is_file()]
if missing:
print(f"[失败] 缺少文件:{missing}")
return 1
problems = check_syntax(paths)
for path in paths:
source = path.read_text(encoding="utf-8")
# ---- import 的目标必须存在,且名字必须真被导出 ----
imported: set[str] = set()
for match in IMPORT_RE.finditer(source):
spec = match.group("src")
if match.group("names"):
for piece in match.group("names").strip("{}").split(","):
piece = piece.strip()
if piece:
imported.add(piece.split(" as ")[-1].strip())
if match.group("default"):
imported.add(match.group("default"))
target = resolve_import(path, spec)
if target is None:
if spec.startswith("."):
problems.append(f"{path.name}: import '{spec}' 指向的文件不存在")
continue
available = exported_names(target)
wanted = set()
if match.group("names"):
for piece in match.group("names").strip("{}").split(","):
piece = piece.strip()
if piece:
wanted.add(piece.split(" as ")[0].strip())
for name in wanted - available:
problems.append(
f"{path.name}: 从 {target.name} 导入了 '{name}',但那边没有导出它"
)
# ---- 用到的全大写常量必须有来源 ----
declared = set(re.findall(r"(?:const|let|var|function|class)\s+([A-Z][A-Z0-9_]{2,})\b", source))
declared |= exported_names(path)
# 先清掉注释 / 字符串 / 模板文字,再排除对象字面量的 key。
stripped = OBJECT_KEY_RE.sub(" ", code_only(source))
for match in CONST_USE_RE.finditer(stripped):
name = match.group(1)
if name in imported or name in declared or name in KNOWN_GLOBALS:
continue
problems.append(f"{path.name}: 使用了 '{name}',但既没 import 也没在本文件声明")
print(f"检查 {len(paths)} 个模块:{', '.join(p.name for p in paths)}")
if problems:
print(f"\n发现 {len(problems)} 个问题:")
for problem in problems:
print(f" · {problem}")
print(
"\n提示:拆分模块时最容易漏 import —— 定义搬走了、使用处还留在原文件,"
"\n 浏览器里表现为页面白板 + ReferenceError,而 Python 测试全绿。"
)
return 1
print("全部通过:语法正确、import 可解析、常量都有来源。")
return 0
if __name__ == "__main__":
sys.exit(main())